Uplink multi-user multiple-input multiple-output (MU-MIMO) is a key technology of the new-generation wireless local area network (WLAN) standard 802.11ax. In the uplink MU-MIMO, an access point (AP) obtains channel estimation of the uplink MU-MIMO by using a high efficiency long training field (HE-LTF) in an uplink packet transmitted by each STA, to demodulate signals from different stations (STA).
FIG. 1 is an exemplary diagram of a training field solution using sub-carrier or sub-band interleaving in the prior art. As shown in FIG. 1, a training field includes 4+1 orthogonal frequency division multiplexing (OFDM) symbols. The training field is divided into two parts according to functions. A first part includes the first four OFDM symbols, that is, an LTF-1, an LTF-2, an LTF-3, and an LTF-4, and a second part includes a fifth OFDM symbol, that is, an additional LTF. Sub-carriers of each OFDM symbol in the first part of the training field each carry a known reference symbol. Four spatial flows sequentially correspond to different sub-carriers or sub-bands in each OFDM symbol, and have locations that do not overlap with those of corresponding sub-carriers or sub-bands of corresponding spatial flows in an adjacent OFDM symbol. A sub-carrier or sub-band corresponding to each spatial flow repeats every four sub-carriers or sub-bands, and after four OFDM symbols, each spatial flow traverses all sub-carriers, but sub-carriers corresponding to the spatial flows are orthogonal to each other. In this way, by using a known reference symbol carried by sub-carriers corresponding to each spatial flow, channel estimation of the corresponding spatial flow on each sub-carrier can be obtained.
In the prior art shown in FIG. 1, the additional LTF in the second part of the training field is a repetition of the LTF-1 in the first part of the training field, and is mainly used to estimate a carrier frequency offset (CFO) corresponding to each spatial flow, so as to compensate for a phase offset caused due to the CFO during channel estimation. However, channel estimation obtained by using the prior art is of low precision, and estimation performance is relatively poor.